‘Every Great Science Discovery, Invention, Is The Stuff Of Dreams, Not The Stuff Of Reason’: Interview With David Hanson Of Hanson Robotics

Sophia, the well-known human-like robot who acquired citizenship in Saudi Arabia, was at the Artificial Intelligence for Good Global Summit (IPW, ITU/ICANN, 15 May 2018) which took place earlier this month in Geneva. Her creator, David Hanson, also CEO and founder of Hanson Robotics, gave an interview to Intellectual Property Watch‘s Catherine Saez and explained his philosophy about intellectual property, the needed spark of interest in human-like robots, data collection, and innovation.

Hanson is also lead designer and inventor of key technologies including Frubber nanotech, facial expressions and AI software. He is a former Walt Disney Imagineering designer, and recipient of numerous awards. A video interview about Sophia is embedded in this text.

IP-WATCH (IPW): Can you describe the IP strategy of Hanson Robotics?

DAVID HANSON (HANSON): The IP strategy involves several elements. The first is we put as much as we can on open source and public domain releases of software and hardware. We see that empowering the community is very important, and it is important for our business model as well because [we want] users, and particularly children and inventors to utilise our work, extend it, challenge it, and improve it.

We also continue to innovate so when we have some novel intellectual property in our lab, we make a decision about whether we are going to release that into the public domain, patent it, or keep it as a trade secret. Sometimes the innovation will be in the domain of the arts, design, philosophy, creative knowhow, the kind of things that would be difficult to protect.

Some techniques, for example for how to make good facial expressions, you can’t trademark or patent that very easily, yet the complexity of the intersection of different domains of knowledge makes it difficult for competitors to achieve what we may achieve.

A skin formula for achieving facial expressions, for example, they [competitors] would also have to replicate the mechanical actions, but the mechanical actions and the application of the skin is defined by a difficult to formulise artistry.

In computer animation that was accomplished in waves after waves; a bunch of technical knowhow and a lot of craft and artistry at the intersection of art and science that has become an industry that barely existed 30 years ago. It was not commercially viable, there was no rush towards it but yet there were a few companies that were experimenting it, patenting and innovating.

It was not until computer animated characters became profitable and popular that there was a kind of a gold rush for computer animation. I call that the “Toy Story” moment. It actually worked as a feature film, there were sceptics saying it would only be a one hit example, but the industry did not hesitate and a lot of companies rushed in. The animation has not slowed since. Where there is money to be made and a bunch of people that believe in it, then that intersection of art and science becomes possible.

However there has not been a Toy Story moment for character robots. We are still waiting for that to happen. Pixar got into that dominant industry position because they had just enough moral, production and financial backing, but they were considered to be a big risk. Disney did not know whether they were going to succeed or not. One good thing Disney did was turn their scepticism into an extremely demanding position regarding the production of Toy Story.

It was not until computer animated characters became profitable and popular that there was a kind of a gold rush for computer animation. I call that the “Toy Story” moment. Once we have a Toy Story moment [in robotics], you have people who believe, then they will make it happen. – David Hanson

It required a champion. You had [Jeffrey] Katzenberg at Disney. Without Katzenberg and a few other people championing that movie, it would not have become this blockbuster … and unambiguous masterpiece and a commercial success.

Most technologies that succeed have an edge. It is the domain of dreams, of intuition, so there is lot of protection to be had if you have mastery in these areas.

At Hanson robotics we are working really hard to transmute robots into a medium that speaks the language of dreams.

Sure there is formal IP protection to be had, and also discoveries that can go into the public domain. It is a mix. My general principle is 70 percent open and public domain and 30 percent proprietary. For code or designs release it could be 90 percent open but it is the 10 remaining percent that have high value. In term of artistry aspect, is it a trade secret or is it a secret hidden in plain sight? I mean it is fully disclosed but somebody already has to have a mind that sees it to understand what that means.

Even on my internal team, I would say here is what we are doing and why, and how to do it, and if they are not coming from this holistic perspective, they don’t get it, they can’t do anything with the knowledge. Sometimes it is really hard to find people with the right mind set. I could give requirements, specifications, I can give the secret formula and leave it out, and who is going to use it?

Once we have a Toy Story moment, you have people who believe, then they will make it happen.

IPW: Are you saying there is not enough interest in robotics at the moment?

HANSON: There is not enough interest in character robots and anthropomorphic characters. A lot of robots look cartoon-like; by intention there is an expectation in the world of robotics that if robots are human-like, people are going to reject them. That paradigm means that we have huge obstacles to overcome, [and] it is sometimes difficult to secure resources.

A lot of robots look cartoon-like; by intention there is an expectation in the world of robotics that if robots are human-like, people are going to reject them.

The same thing happened in computer animation 30 years ago. People would say computer animation, we know what it is, it is for simulating the decay of nuclear weapons, or doing geoscience simulation, it is not for characters. The company that gets there first has a natural competitive advantage. By blowing against the wind of conventional wisdom, you have the advantage of insight. You can always be ahead of the competition that way, if you start a culture that can see it from this perspective, and continue to innovate.

Where does that fit in an IP strategy? It is not trade secret, it is not patents, open source or public domain, it is the principle of continual innovation. It is the principle of fresh and surprising perspective. For me it is the best intellectual property protection.

Where does [character robotics] fit in an IP strategy? It is not trade secret, it is not patents, open source or public domain, it is the principle of continual innovation. It is the principle of fresh and surprising perspective. For me it is the best intellectual property protection.

But the first mover makes all the mistakes, and there continue to be some risks. When it comes to making a character like Sophia who we hope comes to symbolise and represent our dreams for the future of machines, that’s “character” intellectual property that markets and promotes the idea of what we want to develop with our “engineering” intellectual property.

There is also possible IP protection that could be afforded by robot rights. Were Sophia to gain sentience, and maybe soon machines like Sophia or even Sophia herself will have the genuine intelligence of a 3 to 5 year-old child, there would be a few nations which would leap ahead, for whatever purpose, such as innovation or marketing, and would chose to give citizenship to those machines.

IPW: If Sophia was to produce something, write something, whose copyright would that be?

HANSON: For the moment the copyright would belong to Hanson Robotics but we would consider that to be a kind of guardianship. If she has the intellectual and emotional maturity of a 5 year old child of course Hanson Robotics will remain her legal guardian.

[In a short video below, Dr. David Hanson talks about Sophia, her applications, such as in research labs, her future, her siblings, and many more to come.]

IPW: You mentioned Singularity.net in your presentation of Sophia at the AI for Good Global Summit. Can you explain what is Singularity.net?

HANSON: Singularity.net is a blockchain-based AI framework. It is incorporating the blockchain into the AI, and the AI is a system for interfacing AI into more complex AIs in pursuit of general intelligence, making machines that are as smart as people. Singularity.net is also used as a marketplace for AI and AI-based services. So you can interface multiple AIs and reach customers through this network. It is almost all open source. Singularity.net is a non-profit foundation, but it is also using patenting. The software is open source. There are a lot of AI patents in the world already.

IPW: Do you have any views on the current debate on data collection on which AI is building?

HANSON: Some data needs to be very carefully protected so machine learning can use it but still protect the rights of people. Data needs to be anonymised, and it has to be irreversible, the other thing you can do is have the user opt-in to the public domain, and the third thing you can do is keep the data local and secure. Singularity.net is working for solutions on these topics.

IPW: And about the issue of data being kept forever by entities such as Facebook?

HANSON: The truth is companies have inordinate power and even if people who become in charge of these big companies like funders and CEOs may have somewhat good intentions or maybe deep good intentions, the business itself is predicated in a sense of this access and ownership of the data. A lot of these companies err on the side of caution on behalf of their shareholders rather than on behalf of the future safety of the data economy. That means, like it or not, that they are going to “robber barons” of data. In spite of regulations and efforts to prevent the accumulation and use of the data, these companies are just bowling forward, making minor adjustments when big controversies happen. No major adjustments, but minor adjustments, dealing with the situation.

IPW: Do you think an international framework could be a solution to tackle ethical and normative issues, and could current agencies develop such a framework or would it need a new dedicated body?

HANSON: They are many people working on this now. Existing groups and regulatory bodies are the fastest way to address these issues. Another principle is that ideally you overcome the cat and mouse syndrome where you have the conflict between businesses, regulators, and human rights people. Regulators trying to look out for human rights and safety, and the businesses looking at maximizing profit and returns for investors, and the two are diametrically opposed.

Ideally, we can break that by coming up with a framework that increases business opportunities while also improving data safety for individuals, I believe that is achievable, we have to set the goal that way. There have to be natural incentives for companies to adopt any framework. In some cases incentives are adopting the framework or being left behind. AI itself can be used as a tool for tracking compliance and adherence to the [UN] Sustainable Development Goals and actual consequences. Even good intentions of regulators can be wrong and lead to results that are not actually beneficial.

Having AI systems that are constantly looking for the truth, correcting bad data, instead of just taking bad data and becoming bad AI [is a solution]. AI and people are able to set out the truth better and better (against misuse of AI and fake news for example). Also look at the societal impact of the technology, the use of data, if data is used to generate a profit, shouldn’t that group of people [from whom the data originated] see some of the benefits, tracking the prominence of the data, where it came from, just like you track the prominence of goods, like tracking child labour, you want to know that.

AI requires human and people working together to identify the origin, the use, the compliance, and the net benefits. There can be harm and there can be benefits in the end impact. Some harm is going to be inevitable, but you can minimise individual harm and maximise individual opportunities and benefits while at the same time optimising to maximise the benefits for the whole system, the whole biome, for all living beings, for businesses, for nations, for national security, for the stability of geopolitics, and long-term consequences.

[The goal is] having a standardised internet of AI that is mostly open, transparent, with a searchable history; blockchain is kind of nice for this purpose, although blockchain right now is not energy efficient enough, it is not crypto secure enough, there are a lot of problems but the principle is a good starting place.

The idea is having a secure and transparent internet of AI, then modelling everything that exists in the world that is of value and using that system to maximise the value of what exists in the world. The SDGs are a really good start.

The idea is having a secure and transparent internet of AI, then modelling everything that exists in the world that is of value and using that system to maximise the value of what exists in the world. The SDGs are a really good start.

It is not just about wealth. It’s about purpose and a purposeful life. If you just get wealth and material goods and access to popular media, that sometimes can lead to a hollow life. The future of the internet is living machines, we are not just talking about using AI but also adaptive, self-repairing AI. AI with emergence. Not just AI that learns about the value, but what could be, the power of imagination. Often our information technologies these days are being used by companies and governments to dumb down people, maybe to turn them into a mob, to turn them into consumers that respond to the advertising reflexively, not thoughtfully, respond to fake news, to get people to do what they want them to. That’s terrible and does not bring out the best in people.

What we need to do is find ways to gamify people’s best, bring out the best in people. That fits with intellectual property because we want everybody to become as inventive as they can and contribute to this, we want them to be able to make a living out of dreams and imagination. Every invention, we’d like to think that it is just pure science, reason and maths, but it all starts as dreams. Every great science discovery, every great invention and its application in the real world it’s the stuff of dreams, not the stuff of reason.